Int. J. Dev. Biol. 62: 857-864 (2018) https://doi.org/10.1387/ijdb.180210yb

www.intjdevbiol.com

Lung cancer susceptibility genetic variants modulate HOXB2 expression in the lung ALISSON CLEMENCEAU#,1, OLIVIER BOUCHERAT#,2, KIM LANDRY-TRUCHON#,2, MAXIME LAMONTAGNE1, SABRINA BIARDEL1, PHILIPPE JOUBERT1,3, STÉPHANE GOBEIL4,5, BLANDINE SECCO1, MATHIEU LAPLANTE1,3, MATHIEU MORISSETTE1,3, MA’EN OBEIDAT7, WIM TIMENS8, LUCIE JEANNOTTE*,2,3,6 and YOHAN BOSSÉ*,1,3,5

1Institut Universitaire de Cardiologie et de Pneumologie de Québec, Université Laval, Québec, Canada, 2CRCHU Québec- Université Laval, Québec, Canada, 3Centre de Recherche sur le Cancer de l'Université Laval, Québec, Canada, 4CRCHU de Québec, CHUL, Québec, Canada, 5Department of Molecular Medicine, Université Laval, Québec, Canada, 6Department of Molecular Biology, Medical Biochemistry and Pathology, Université Laval, Québec, Canada, 7The University of British Columbia, Centre for Heart Lung Innovation, St Paul’s Hospital, Vancouver, BC, Canada, 8Department of Pathology and Medical Biology, University of Groningen, University Medical Center Groningen, GRIAC Research Institute, Groningen, The Netherlands

ABSTRACT The HOX are transcription factors that are expressed in coordinated spatio- temporal patterns to ensure normal development. Ectopic expression may instead lead to the development and progression of tumors. Genetic polymorphisms in the regions of four HOX clusters were tested for association with lung cancer in 420 cases and 3,151 controls. The effect of these variants on lung (expression quantitative trait loci, eQTL) was tested in a discovery set of 409 non-tumor lung samples and validated in two lung eQTL replication sets (n = 287 and 342). The expression levels of HOXB2 were evaluated at the mRNA and levels by quantitative real-time PCR and immunohistochemistry in paired tumor and non-tumor lung tissue samples. The most significant SNP associated with lung cancer in the HOXB cluster was rs10853100 located upstream of the HOXB cluster. HOXB2 was the top eQTL-regulated gene with several polymorphisms associated with its mRNA expression levels in lung tissue. This includes the lung cancer SNP rs10853100 that was significantly associated withHOXB2 expression (P=3.39E-7). In the lung eQTL discovery and replication sets, the lung cancer risk allele (T) for rs10853100 was associated with lower HOXB2 expression levels. In paired normal-tumor samples, HOXB2 mRNA and protein levels were significantly reduced in tumors when compared to non-tumor lung tissues. Genetic variants in the HOXB cluster may confer susceptibility to lung cancer by modulating the expression of HOXB2 in the lung.

KEY WORDS: HOXB2, lung adenocarcinoma, lung eQTL

Introduction tern formation in animals. HOX genes specify embryonic regional identity and control morphogenesis (Krumlauf, 1994). In human Aberrant expression of molecular regulators originally described and mouse, 39 HOX genes are ordered in four clusters located as critical factors of mammalian embryogenesis may contribute on different (Table 1). Their position within a cluster to various pathologies, including cancer (Bonner et al., 2004). correlates with their respective spatiotemporal expression during Among the genes implicated in normal development, HOX genes act as master regulators (Bhatlekar et al., 2014). They encode Abbreviations used in this paper: eQTL, expression quantitative trait loci; FDR, false an evolutionarily conserved family of transcription factors, which discovery rate; LD, linkage disequilibrium; NSCLC, non-small cell lung cancer; regulate the expression of downstream effectors that direct pat- SNP, single nucleotide polymorphism.

*Address correspondence to: Joint senior authors: Yohan Bossé. Institut universitaire de cardiologie et de pneumologie de Québec, Pavillon Marguerite-d’Youville, Y2106, 2725 chemin Sainte-Foy, Québec, QC, Canada, G1V 4G5. Tel: (418) 656-8711 ext. 3725. Fax: (418) 656-4940. E-mail: [email protected] - https://orcid.org/0000-0002-3067-3711 or Lucie Jeannotte. Centre de recherche sur le cancer de l’Université Laval, CRCHU Québec-Université Laval, L’Hôtel- Dieu de Québec, 9, rue McMahon, Québec, QC, Canada G1R 3S3. Tel: (418) 525-4444 ext. 15552. Fax: (418) 691-5439. E-mail: [email protected] - https://orcid.org/0000-0001-5578-420X - #Note: Joint first authors.

Supplementary Material (two tables) for this paper is available at: https://doi.org/10.1387/ijdb.180210yb

Submitted: 7 July, 2018; Accepted: 13 September, 2018.

ISSN: Online 1696-3547, Print 0214-6282 © 2018 UPV/EHU Press Printed in Spain 858 A. Clemenceau et al.

Fig. 1. Lung expression quantitative trait loci (eQTL) in the HOXB gene cluster. The y-axis represents eQTL P value in -log10 scale. The positions of the genes are illustrated at the bottom. Each dot represents the association (P value) between a SNP and the expression of a specific gene that is color coded, e.g. red dots represent the association of SNPs with the mRNA expression levels of HOXB2.

embryogenesis. This organization is essential for the specific human non-small cell lung cancer (NSCLC) correlates with tumor regulation and function of each and hence for the cor- node metastasis and poor patient survival because gain of HOXC8 rect patterning of the embryo (Tschopp et al., 2009). Accordingly, expression promotes proliferation and migration via the activation HOX genes are expressed in defined but overlapping domains of transforming growth factor beta 1 (TGFB1) expression (Liu et along the developing axes. al., 2018). The expression of HOXB2 was also demonstrated as In the lung, several HOX genes, mainly from the 3’ end of the an adverse prognostic factor for stage I lung adenocarcinoma HOXA and HOXB clusters, are expressed along the respiratory through its role in promoting invasion of lung cancer cells (Inamura tract, each with a distinct proximal-distal distribution (Herriges et et al., 2008, Inamura et al., 2007). In contrast, reduced HOXA5 al., 2012). Mouse genetic studies have shown that only Hoxa5 expression is reported to be an adverse prognosis indicator for single mutant mice display respiratory tract defects, which cause NSCLC as HOXA5 restrains NSCLC cell proliferation by activating respiratory distress and death at birth (Aubin et al., 1997, Landry- the expression of cyclin-dependent kinase inhibitor p21 (Zhang et Truchon et al., 2017). This demonstrates the functional redundancy al., 2015). However, other studies on NSCLC reported divergent between HOX genes due to expression overlap (Boucherat et data, and expression levels of HOXA1, HOXA5, HOXA10 and al., 2013). HOXC6 genes were found to be significantly increased in lung HOX genes are involved in multiple cellular processes, in- squamous cell carcinoma (Abe et al., 2006). Higher expression cluding cell differentiation, proliferation and survival, which once of HOXA5, HOXA10 and HOXB2 was also reported in lung ad- deregulated can lead to cancer (Abate-Shen, 2002). Inappropriate enocarcinoma when compared to non-cancerous tissues (Abe HOX gene expression is observed in multiple cancers supporting et al., 2006, Inamura et al., 2007). These discrepancies make the notion that HOX deregulation is involved in the development it challenging to decipher the role of HOX genes in lung cancer. and progression of tumors. For instance, HOXC8 upregulation in In the present study, we analyzed genetic polymorphisms in TABLE 1

HOX GENES

Gene clusters Boundaries (bp) Genes HOXA 7 27,122,614-27,249,725 HOXA1, HOXA2, HOXA3, HOXA4, HOXA5, HOXA6, HOXA7, HOXA9, HOXA10, HOXA11, HOXA13 HOXB 17 46,596,807-46,816,111 HOXB1, HOXB2, HOXB3, HOXB4, HOXB5, HOXB6, HOXB7, HOXB8, HOXB9, HOXB13 HOXC 12 54,322,576-54,459,814 HOXC4, HOXC5, HOXC6, HOXC8, HOXC9, HOXC10, HOXC11, HOXC12, HOXC13 HOXD 2 176,954,530-177,065,635 HOXD1, HOXD3, HOXD4, HOXD8, HOXD9, HOXD10, HOXD11, HOXD12, HOXD13 HOX genes and lung cancer 859

TABLE 2 the four HOX gene clusters for association with lung cancer. Gene expression levels of HOX genes were evaluated in non-tumor SNPs ASSOCIATED WITH LUNG CANCER lung tissue and eQTL analyses were performed to identify genetic WITHIN THE HOX GENE CLUSTERS polymorphisms associated with mRNA expression levels of HOX

Frequency Frequency genes in the lung. Genes regulated by lung cancer-associated Cluster SNP Tested allele cases controls P value polymorphisms were further characterized in paired tumor and HOXA rs10228276 G 0.20 0.17 0.0101 non-tumor lung tissue samples at the mRNA and protein levels. rs13243033 T 0.20 0.17 0.0104 rs6949451 T 0.52 0.47 0.0122 Results rs3779456 C 0.38 0.43 0.0138 rs6970537 A 0.38 0.43 0.0201 rs2189239 T 0.10 0.08 0.0303 Association between genetic polymorphisms in the HOX gene rs2428432 T 0.29 0.25 0.0342 clusters and lung cancer rs10233387 A 0.49 0.45 0.0364 104 SNPs common to both cases (n = 420) and controls (n = HOXB rs10853100 T 0.54 0.49 0.0063 3,151) located within the HOX clusters were tested for association rs11079833 T 0.04 0.03 0.0238 with lung cancer (Supplementary Table 1). SNPs passing a liberal rs733920 T 0.12 0.15 0.0394 P value of 0.05 are indicated in Table 2. The SNP most strongly rs890431 T 0.07 0.09 0.0464 rs15689 C 0.29 0.26 0.0479 associated with lung cancer, namely rs2067554, is located on HOXC rs2067554 T 0.43 0.37 0.0031 human chromosome 12 in the HOXC cluster. This SNP is located rs1386016 A 0.38 0.42 0.0232 near the 3’UTR of HOXC4, the most upstream HOX gene in this HOXD rs3754983 A 0.09 0.13 0.0050 cluster. The second significant SNP associated with lung cancer in the cluster (rs1386016) is in modest LD with rs2067554 (r2 = TABLE 3 0.37). A single SNP in the HOXD cluster on chromosome 2 was CLINICAL CHARACTERISTICS OF THE LUNG eQTL DISCOVERY associated with lung cancer (rs3754983). This SNP is located at AND REPLICATION SETS the 5’ end of the HOXD cluster in the promoter region of HOXD13. Five SNPs were significantly associated with lung cancer in the Discovery set Replication 1 Replication 2 (n = 409) (UBC, n = 287) (Groningen, n = 342) HOXB cluster on . The most significant was Sex (male:female) 229:180 155:132 182:160 rs10853100 located upstream of the HOXB cluster. Finally, 8 Age (years)* 63.3 ± 9.9 61.6 ± 13.2 51.5 ± 15.6 SNPs within the HOXA cluster on chromosome 7 were associated Smoking status, n (%) with lung cancer. These SNPs are located in four LD blocks: 1) Never smokers 36 (9) 26 (9) 100 (29) rs2428432 located in the HOXA2-HOXA3 intergenic region, 2) Former smokers 283 (69) 163 (57) 185 (54) rs6949451, rs3779456 and rs6970537 located in the HOXA10 Current smokers 90 (22) 98 (34) 57 (17) gene, 3) rs2189239 and rs10233387 located within or near the Primary diagnostic, n (%) Adenocarcinoma 235 (57) 78 (27) 36 (11) HOXA13 gene, and 4) rs10228276 and rs13243033 located in Squamous cell carcinoma 104 (25) 82 (29) 51 (15) the most downstream region of the HOXA cluster. Others** 70 (17) 127 (44) 255 (75) *mean ± sd Lung eQTLs in the HOX gene clusters **Others include patients that underwent lung surgery for other histological subtypes of lung cancer, COPD, alpha-1 antitrypsin deficiency, cystic fibrosis, idiopathic pulmonary fibrosis, bronchiectasis, Table 3 shows the clinical characteristics of the final 409 subjects pulmonary hypertension, and sarcoidosis. used in the eQTL analysis. Only one significant lung eQTL was

TABLE 4

LUNG eQTLs DETECTED IN THE DISCOVERY SET FOR THE HOX CLUSTERS

HOX cluster Gene # of eQTL-SNP Range P value Range FDR HOXA HOXA9 1 1.49E-5 4.01E-3

HOXB HOXB2 173 4.96E-11 to 1.15E-2 6.19E-9 to 3.81E-2 HOXB4 23 5.90E-6 to 1.82E-3 9.54E-4 to 4.57E-2 HOXB7 113 2.98E-5 to 8.87E-3 3.12E-3 to 4.52E-2

HOXC HOXC6 42 4.85E-5 to 5.36E-3 9.45E-3 to 4.38E-2

HOXD HOXD3 40 1.65E-5 to 6.55E-3 1.20E-3 to 3.75E-2 HOXD1 25 7.30E-4 to 3.88E-3 1.33E-2 to 3.55E-2

Fig. 2. HOXB2 gene expression levels in lungs according to genotype groups for SNP rs10853100 in the discovery set. The y axis represents gene expression levels in the lung (n = 409). The x axis represents the three groups according to their genotypes. The number of subjects per groups is indicated in parentheses. The right y-axis shows the proportion of the gene expression variance explained by the SNP (black bar). 860 A. Clemenceau et al.

Fig. 3. Gene expression levels of the HOXB2 gene in lungs according to genotype groups for SNP rs10853100 in the replication set from UBC (left panel) and the replication set from Groningen (right panel). The y axis represents HOXB2 expression levels in the lung (n = 287 and 342 for UBC and Groningen, respectively). The x axis represents the three groups ac- cording to their genotypes. The number of subjects per groups is indicated in parentheses. The right y-axis shows the proportion of the gene expression vari- ance explained by the SNP (black bar). found in the HOXA cluster at a FDR of 5% (Table 4). In contrast, the eQTL results of the HOX genes residing in this region. A total of the three other clusters were characterized by many lung eQTLs. 173 SNPs regulated the expression of HOXB2, 23 for HOXB4, and Supplementary Table 2 provides the list of significant eQTLs in 113 for HOXB7. Many SNPs regulating the expression of HOXB2 the four clusters. In the HOXC cluster, SNPs were associated with also regulated HOXB7 expression, suggesting co-regulation of the the expression of HOXC6 (Table 4). For the HOXD cluster, SNPs two genes. However, the association signals were predominantly were associated with the expression of HOXD3 and HOXD1 (Table higher for HOXB2. Interestingly, the SNP most strongly associ- 4). The highest number of eQTLs detected in this study were ated with lung cancer in this region (rs10853100; Table 2) was located in the HOXB complex (Table 4). The significant levels of significantly associated with the expression of HOXB2 (Fig. 2). eQTLs were also orders of magnitude higher in the HOXB cluster The lung cancer risk allele (T) for this polymorphism was associ- compared to the other clusters (P=10-9 vs P=10-3). We have thus ated with lower HOXB2 RNA expression levels in lung (Fig. 2). focused subsequent analyses on the HOXB cluster. Fig. 1 shows Two replications sets were used to confirm the lung eQTL rs10853100-HOXB2. Table 3 shows the clinical characteristics of each replication set. This eQTL was reproduced in both UBC and Groningen replication sets, suggesting that the risk allele in the HOXB cluster on chromosome 17 mediates its effect on lung cancer development through regulation of HOXB2 in the lung (Fig. 3).

Decreased HOXB2 expression in lung tumor The association between rs10853100 and expression levels of HOXB2 was further validated by qPCR analyses in tumor and non- tumor tissues from a different set of patients including 16 individuals with genotype CC and 16 with genotype TT. The clinical char- acteristics of patients by genotyping groups for the qPCR experiment are shown in Table 5. Fig. 4 shows HOXB2 expression values grouped either by tissue types or genotype groups. HOXB2 levels were significantly lowered in tumors compared to non-tumor lung tissues. In contrast, no difference was observed between the TT and CC genotypes in non-tumor tissues as well as in tumor specimens. Note that the pathological stage Fig. 4. HOXB2 expression levels according to tissue types and genotype groups. The y axis was not associated with HOXB2 expression represents HOXB2 expression levels normalized to GAPDH. The number of individuals is indicated in these patients (Kruskal-Wallis chi-squared in parentheses at the bottom of each bar. = 2.0638, P = 0.36). HOX genes and lung cancer 861

Fig. 5. HOXB2 protein expression is reduced in lung adenocarcinoma. IHC experiments against HOXB2 protein was performed on sections of lung tumor tissues and their associated control from 9 different patients. HOXB2 staining was weaker in the majority of the tumor tissues (A’-I’) when compared to controls (A-I). The pathological stage is indicated at the bottom left corner of tumor panels. Scale bar, 100 mm.

We also performed IHC experiments on lung adenocarcinoma compared between patients and controls. The strongest evidences and non-tumor specimens for HOXB2 protein expression. Consis- of genetic associations identified were for SNP rs2067554 located tent with the qPCR data, decreased HOXB2 protein expression near the 3’UTR of HOXC4 and SNP rs10853100 located upstream was observed in the vast majority of the tumor specimens (Fig. 5). of the HOXB cluster. Using a large-scale lung eQTL dataset, HOXB2 was found as the most statistically significant eQTL-regulated Discussion gene. Interestingly, the disease-associated SNP rs10853100 was associated with the expression of HOXB2 and direction of effect In this study, the variants in the HOX genes’ family were investi- was that the lung cancer risk allele (T) correlated with lowered gated for their potential role in the susceptibility and pathogenesis of expression of HOXB2. This direction of effect was confirmed in two lung cancer, the leading cause of cancer death. Allele frequencies large and independent lung eQTL datasets. The mRNA expression of germline variants located in the four HOX gene clusters were and protein levels of HOXB2 were found to be reduced in tumors 862 A. Clemenceau et al.

TABLE 5 and cervical cancers (Gonzalez-Herrera et al., 2015, Segara et al., 2005). In contrast, expression was decreased in breast cancer CLINICAL CHARACTERISTICS OF PATIENTS cell lines and acute myeloid leukemia, suggesting that it can act FOR THE VALIDATION STEP BY qPCR ACCORDING as a tumor suppressor (Boimel et al., 2011, Lindblad et al., 2015). TO GENOTYPING GROUPS FOR SNP RS10853100 This supports the notion that aberrant expression of HOXB2 gene

CC TT is detrimental to normal cell growth. Gender (male:female) 7:9 8:8 In the qPCR experiment, no difference in HOXB2 expression Age (years)* 63.6 ± 8.8 59.8 ± 11.2 was observed between the TT and CC genotypes in non-tumor Smoking status, n (%) tissues as well as in tumor specimens. Lower expression was ex- Never smokers 1 (6) 0 (0) pected for TT, at least in non-tumor tissues, considering the results Former smokers 13 (81) 9 (56) of the three independent eQTL datasets (n=409 Laval, 287 UBC, Current smokers 2 (13) 7 (44) Primary diagnostic, n (%) and 342 Groningen). We believe that the sample size was not Adenocarcinoma 16 (100) 16 (100) sufficient in the qPCR experiment (n=32 patients) to demonstrate Squamous cell carcinoma 0 (0) 0 (0) the effect of the genotype. However, this sample size was enough Others 0 (0) 0 (0) to demonstrate lower expression of HOXB2 in tumors compared *mean ± sd to non-tumor lung tissues. compared to paired non-tumor lung tissues. Taken together, these Our study has limitations. First, the case-control genetic as- results suggest that common variants in the HOXB cluster confer sociation study was conducted using a shared control group not susceptibility to lung cancer through down-regulation of the HOXB2 phenotyped for lung cancer. Accordingly, some of the controls are mRNA expression levels in the lung. likely to develop lung cancer in the future and may result is some The mRNA expression level of HOXB2 has been identified misclassification and dilution of the signal. Second, the genetic as a prognostic indicator for stage I lung adenocarcinomas in a association results were not replicated in an independent dataset. Japanese population (Inamura et al., 2007). The group with higher Instead, lung eQTL datasets were used to mechanically explain expression (n=15) had worse disease-specific survival compared the genetic association findings into gene expression in the lung. to the group with lower expression (n=37). In a subset of these Third, no functional work (in vitro and/or in vivo) was performed cases with normal lung tissue samples available, HOXB2 expres- to demonstrate the causal role of HOXB2 and disease-associated sion was higher in tumor relative to normal tissues. In the same SNPs in lung cancer. Finally, the genetic association study and lung study, HOXB2 expression in normal lung tissues was higher in eQTL datasets were obtained from patients of European ancestry metastatic cancer patients. A follow-up study from the same group only. The results may thus not translate into other ethnic groups. proposed that HOXB2 promotes NSCLC cells invasion through the In summary, genetic variants in the HOX gene clusters show up-regulation of metastasis-related genes, including genes coding association with lung cancer. One of the top disease-associated for glycoproteins and signaling molecules (Inamura et al., 2008). variant was also associated with the expression levels of HOXB2 Together these results suggested that higher HOXB2 expression in non-tumor lung tissues. The lung cancer risk allele (T) was as- favor tumorigenesis and/or invasion. In contrast, in the current sociated with lower expression of HOXB2 in three independent study, we demonstrated reduced mRNA and protein expression datasets (discovery and two replication sets). Consistent with this of HOXB2 in adenocarcinomas compared to adjacent normal lung direction of effect, the mRNA and protein levels of HOXB2 were tissues. Although this seems contradictory, similar observations lower in tumors compared to adjacent normal lung tissues. These were made for a positive regulator of HOX, namely PREP1 (Penkov results highlight HOXB2 as a candidate gene for lung cancer and et al., 2013). PREP1 was shown to be absent or down-regulated suggest that down-regulation of this gene as a potential mechanism in human tumors compared to normal tissues (Longobardi et al., involved in lung cancer development. Validation of these findings 2010), as we observed for HOXB2. On the other hand, it was and functional studies are needed to establish the causality of found to be a negative prognostic factor in patients with stage HOXB2 for lung cancer. I NSCLC (Risolino et al., 2014), as observed for HOXB2 in the Japanese population described above. PREP1 is recognized as Materials and Methods a tumor suppressor that inhibits tumor growth, but at the same time an inducer of the epithelial-mesenchymal transition and a Discovery set prometastatic (Risolino et al., 2014). It is thus Lung specimens (n = 420) from lung cancer patients were obtained possible that both up- and down-regulation are detrimental and a from the biobank of the Institut universitaire de cardiologie et de pneu- fine balance in PREP1 and also HOXB2 expression is required. mologie de Québec (IUCPQ). Corresponding clinical data were obtained In our recent transcriptomic study (Bossé et al., 2017), HOXB2 at the time of tissue collection and available in a local database including was not differentially expressed in stage I lung adenocarcinomas demographic variables and detailed pathology report. Written informed compared to adjacent non-tumor lung samples collected at 0, 2, consent was obtained from all subjects. The study was approved by the local ethics committee. 4, and 6 cm away from the tumor. In addition, HOXB2 expression levels in non-tumor sites were relatively stable across the differ- Replication sets ent distances from the tumor. Larger sample size studies will be Significant lung eQTL identified in the discovery set were replicated needed to determine whether HOXB2 is up- or down-regulated in two independent cohorts. The first replication set consists of patients in lung tumor. recruited at the University of British Columbia (UBC; Vancouver, Canada). Changes in HOXB2 expression were previously reported in At UBC, the majority of samples were from patients undergoing resection several types of cancer. HOXB2 was more expressed in pancreatic of small peripheral lung lesions. Additional samples were from autopsy HOX genes and lung cancer 863 and at the time of lung transplantation. Patients from UBC provided written Lung eQTL analyses informed consent and the study was approved by the ethics committee RNA expression traits were adjusted for age, sex, and smoking status at the UBC-Providence Health Care Research Institute Ethics Board. using robust residuals obtained with the rlm function in the R statistical The second replication set was recruited at the University of Groningen package MASS. Residuals values deviating from the median by more than (Groningen, The Netherlands). At Groningen, the lung specimens were three standard deviations were filtered as outliers. Association tests between obtained at surgery from patients with various lung diseases, including adjusted expression traits and SNPs were performed using quantitative patients undergoing therapeutic resection for lung tumors, harvested from association tests implemented in PLINK (Purcell et al., 2007). Significant a site distant from the tumor, and lung transplantation. The study protocol eQTLs were those passing the 5% false discovery rate (FDR) thresh- was consistent with the Research Code of the University Medical Center old (Benjamini and Hochberg, 1995) in the discovery set. P values lower Groningen and Dutch national ethical and professional guidelines (“Code than 0.05 were considered statistically significant in the replication sets. of conduct; Dutch federation of biomedical scientific societies”; http://www. federa.org). The final analyses were performed with 287 and 342 lung qPCR specimens from UBC and Groningen, respectively. Expression values (½ΔΔCT) deviating by more than three standard errors from the mean were considered outliers and removed from analysis. Gene Gene expression and genotyping microarrays expression values for HOXB2 were analyzed by tissue types (non-tumor Gene expression profiling and genotyping were described previous- vs tumor) and genotypes (CC vs TT). Two-sided t-tests and paired t-tests ly (Hao et al., 2012, Lamontagne et al., 2013). Briefly a non-neoplastic pul- were used to compare genotyping groups and tissue types, respectively. monary parenchyma sample was harvested at surgery. RNA was extracted from the lung specimens and then hybridized to a custom Affymetrix array Acknowledgments (GEO platform GPL10379). Expression values were extracted using the The authors would like to thank the research staff at the Quebec Respi- Robust Multichip Average (RMA) method (Irizarry et al., 2003) and probe ratory Health Network Biobank, site IUCPQ for their valuable assistance. sets testing HOX genes were considered in this study. Genotyping was This study was supported by the Fondation de l’Institut universitaire de carried using the Illumina Human1M-Duo BeadChip. Single nucleotide cardiologie et de pneumologie de Québec (to Y.B.), the Quebec Respiratory polymorphism located 10 kb up- and downstream of the four HOX clusters Health Network (to Y.B.), Centre de Recherche sur le Cancer de l’Université were analyzed in this study (Table 1). Laval (to L.J. and Y.B.), Canadian Institutes of Health Research (MOP- 15139 to L.J. and MOP-123369 to Y.B.), and Cancer Research Society Quantitative real-time PCR A subset of patients with lung adenocarcinoma enrolled in the lung (to L.J. and to Y.B.). Y.B. holds a Canada Research Chair in Genomics of eQTL mapping study was selected for validation by quantitative real-time Heart and Lung Diseases. PCR (qPCR). They were selected based on tissue availability and using a random stratified strategy based on genotypes for the single nucleotide References polymorphism (SNP) rs10853100. Tumor and non-tumor lung specimens from sixteen individuals with genotype CC and sixteen with genotype TT were ABATE-SHEN C. (2002). Deregulated gene expression in cancer: cause evaluated. RNA from tumor and non-tumor specimens was extracted from 30 or consequence? Nat Rev Cancer 2: 777-785. mg of frozen tissues using the RNeasy Plus Universal Mini kit (Qiagen, MD, ABE M, HAMADA J, TAKAHASHI O, TAKAHASHI Y, TADA M, MIYAMOTO M, MORI- USA). RNA concentration and purity were assayed by UV 260/280 nm ratio KAWA T, KONDO S and MORIUCHI T. (2006). Disordered expression of HOX on a NanoVue spectrophotometer (GE Healthcare, NJ, USA). The quality genes in human non-small cell lung cancer. Oncol Rep 15: 797-802. of each RNA sample was then evaluated by automated electrophoresis on AUBIN J, LEMIEUX M, TREMBLAY M, BÉRARD J and JEANNOTTE L. (1997). the Experion system (Bio-Rad, CA, USA). cDNA was synthesized with the Early postnatal lethality in Hoxa-5 mutant mice is attributable to respiratory tract Superscript II Reverse Transcriptase (Invitrogen) using random primers. defects. Dev Biol 192: 432-445. Quantitative PCR (qPCR) was performed with Power SYBR Green PCR BENJAMINI Y and HOCHBERG Y. (1995). Controlling the false discovery rate: a practi- Master Mix (Applied Biosystems, Foster City, CA USA) and a thermal cycler cal and powerful approach to multiple testing. J Royal Stat Soc Ser B 57: 289-300. ABI PRISM 7000. Samples were analyzed in triplicate. The primer sequences BHATLEKAR S, FIELDS J Z and BOMAN B M. (2014). HOX genes and their role in used for the HOXB2 gene were: forward-5’TCCTTCCCGACCTCAACTTC3’, the development of human cancers. J Mol Med (Berl) 92: 811-823. reverse-5’GAGAACCCTGTAGGCTAGGG3’. The GAPDH was used as ref- BOIMEL P J, CRUZ C and SEGALL J E. (2011). A functional in vivo screen for regu- erence gene using the primers: forward-5’GGTCGGAGTCAACGGATTTG3’ lators of tumor progression identifies HOXB2 as a regulator of tumor growth in and reverse-5’GGAAGATGGTGATGGGATTTC3’. breast cancer. Genomics 98: 164-172. BONNER A E, LEMON W J, DEVEREUX T R, LUBET R A and YOU M. (2004). Immunohistochemistry analysis Molecular profiling of mouse lung tumors: association with tumor progression, Lung adenocarcinoma and non-tumor tissues obtained from surgeries lung development, and human lung adenocarcinomas. Oncogene 23: 1166-1176. were fixed in 10% formaldehyde, embedded in paraffin and sectioned at 4 BOSSÉ Y, SAZONOVA O, GAUDREAULT N, BASTIEN N, CONTI M, PAGE S, m m. Immunohistochemistry (IHC) assay was performed as described (Gen- TRAHAN S, COUTURE C and JOUBERT P. (2017). Transcriptomic microenviron- dronneau et al., 2012). The primary HOXB2 antibody was a mouse ment of lung adenocarcinoma. Cancer Epidemiol Biomarkers Prev 26: 389-396. monoclonal antibody against the human HOXB2 protein (1/100 dilution, BOUCHERAT O, MONTARON S, BERUBE-SIMARD F A, AUBIN J, PHILIPPIDOU Developmental Studies Hybridoma Bank). The secondary antibody was a P, WELLIK D M, DASEN J S and JEANNOTTE L. (2013). Partial functional re- goat anti-mouse biotinylated antibody (1/250, Jackson Lab). Slides were dundancy between Hoxa5 and Hoxb5 paralog genes during lung morphogenesis. counterstained with hematoxylin. Am J Physiol Lung Cell Mol Physiol 304: L817-830. GENDRONNEAU G, BOUCHERAT O, AUBIN J, LEMIEUX M and JEANNOTTE L. Statistical analyses (2012). The loss of Hoxa5 function causes estrous acyclicity and ovarian epithelial inclusion cysts. Endocrinology 153: 1484-1497. Genetic association GONZALEZ-HERRERA A, SALGADO-BERNABE M, VELAZQUEZ-VELAZQUEZ Genotypes from the lung cancer cases from the discovery set (n=420) C, SALCEDO-VARGAS M, ANDRADE-MANZANO A, AVILA-MORENO F and were compared to 3,151 publicly available controls taken from the Illumina® PINA-SANCHEZ P. (2015). Increased expression of HOXB2 and HOXB13 iControlDB. Genetic association tests comparing allele frequencies between is associated with HPV infection and cervical cancer progression. Asian Pac J cases and controls were performed with PLINK (Purcell et al., 2007). Cancer Prev 16: 1349-1353. 864 A. Clemenceau et al.

HAO K, BOSSÉ Y, NICKLE D C, PARE P D, POSTMA D S, LAVIOLETTE M, SAND- up-regulation of TGFbeta1 in non-small cell lung cancer. Oncogenesis 7: 1. FORD A, HACKETT T L, DALEY D, HOGG J C et al., (2012). Lung eQTLs to LONGOBARDI E, IOTTI G, DI ROSA P, MEJETTA S, BIANCHI F, FERNANDEZ- Help Reveal the Molecular Underpinnings of Asthma. PLoS Genet 8: e1003029. DIAZ L C, MICALI N, NUCIFORO P, LENTI E, PONZONI M et al., (2010). Prep1 HERRIGES J C, YI L, HINES E A, HARVEY J F, XU G, GRAY P A, MA Q and SUN (pKnox1)-deficiency leads to spontaneous tumor development in mice and ac- X. (2012). Genome-scale study of transcription factor expression in the branching celerates EmuMyc lymphomagenesis: a tumor suppressor role for Prep1. Mol mouse lung. Dev Dyn 241: 1432-1453. Oncol 4: 126-134. INAMURA K, TOGASHI Y, NINOMIYA H, SHIMOJI T, NODA T and ISHIKAWA Y. PENKOV D, MATEOS SAN MARTIN D, FERNANDEZ-DIAZ L C, ROSSELLO C A, (2008). HOXB2, an adverse prognostic indicator for stage I lung adenocarcinomas, TORROJA C, SANCHEZ-CABO F, WARNATZ H J, SULTAN M, YASPO M L, promotes invasion by transcriptional regulation of metastasis-related genes in GABRIELI A et al., (2013). Analysis of the DNA-binding profile and function of HOP-62 non-small cell lung cancer cells. Anticancer Res 28: 2121-2127. TALE homeoproteins reveals their specialization and specific interactions with INAMURA K, TOGASHI Y, OKUI M, NINOMIYA H, HIRAMATSU M, SATOH Y, OKU- Hox genes/proteins. Cell Rep 3: 1321-1333. MURA S, NAKAGAWA K, SHIMOJI T, NODA T et al., (2007). HOXB2 as a novel PURCELL S, NEALE B, TODD-BROWN K, THOMAS L, FERREIRA M A, BENDER prognostic indicator for stage I lung adenocarcinomas. J Thorac Oncol 2: 802-807. D, MALLER J, SKLAR P, DE BAKKER P I, DALY M J et al., (2007). PLINK: a IRIZARRY R A, HOBBS B, COLLIN F, BEAZER-BARCLAY Y D, ANTONELLIS K J, tool set for whole-genome association and population-based linkage analyses. SCHERF U and SPEED T P. (2003). Exploration, normalization, and summaries Am J Hum Genet 81: 559-575. of high density oligonucleotide array probe level data. Biostatistics 4: 249-264. RISOLINO M, MANDIA N, IAVARONE F, DARDAEI L, LONGOBARDI E, FERNANDEZ KRUMLAUF R. (1994). Hox genes in vertebrate development. Cell 78: 191-201. S, TALOTTA F, BIANCHI F, PISATI F, SPAGGIARI L et al., (2014). Transcription LAMONTAGNE M, COUTURE C, POSTMA D S, TIMENS W, SIN D D, PARÉ P D, factor PREP1 induces EMT and metastasis by controlling the TGF-beta-SMAD3 HOGG J C, NICKLE D, LAVIOLETTE M and BOSSÉ Y. (2013). Refining sus- pathway in non-small cell lung adenocarcinoma. Proc Natl Acad Sci USA 111: ceptibility loci of chronic obstructive pulmonary disease with lung eQTLs. PLoS E3775-3784. One 8: e70220. SEGARA D, BIANKIN A V, KENCH J G, LANGUSCH C C, DAWSON A C, SKALICKY LANDRY-TRUCHON K, HOUDE N, BOUCHERAT O, JONCAS F H, DASEN J S, D A, GOTLEY D C, COLEMAN M J, SUTHERLAND R L and HENSHALL S M. PHILIPPIDOU P, MANSFIELD J H and JEANNOTTE L. (2017). HOXA5 plays tissue- (2005). Expression of HOXB2, a retinoic acid signaling target in pancreatic cancer specific roles in the developing respiratory system.Development 144: 3547-3561. and pancreatic intraepithelial neoplasia. Clin Cancer Res 11: 3587-3596. LINDBLAD O, CHOUGULE R A, MOHARRAM S A, KABIR N N, SUN J, KAZI J U TSCHOPP P, TARCHINI B, SPITZ F, ZAKANY J and DUBOULE D. (2009). Uncoupling and RONNSTRAND L. (2015). The role of HOXB2 and HOXB3 in acute myeloid time and space in the collinear regulation of Hox genes. PLoS Genet 5: e1000398. leukemia. Biochem Biophys Res Commun 467: 742-747. ZHANG M L, NIE F Q, SUN M, XIA R, XIE M, LU K H and LI W. (2015). HOXA5 LIU H, ZHANG M, XU S, ZHANG J, ZOU J, YANG C, ZHANG Y, GONG C, KAI Y and indicates poor prognosis and suppresses cell proliferation by regulating p21 LI Y. (2018). HOXC8 promotes proliferation and migration through transcriptional expression in non small cell lung cancer. Tumour Biol 36: 3521-3531. Further Related Reading, published previously in the Int. J. Dev. Biol.

Cell death in cancer therapy of lung adenocarcinoma Anna Zagryazhskaya, Katarina Gyuraszova and Boris Zhivotovsky Int. J. Dev. Biol. (2015) 59: 119-129 https://doi.org/10.1387/ijdb.150044bz

Epithelial-mesenchymal interactions: a fundamental Developmental Biology mechanism Domenico Ribatti and Marcello Santoiemma Int. J. Dev. Biol. (2014) 58: 303-306 https://doi.org/10.1387/ijdb.140143dr

Homeobox genes from clusters A and B demonstrate characteristics of temporal colinearity and differential restrictions in spatial expression domains in the branching mouse lung R Mollard and M Dziadek Int. J. Dev. Biol. (1997) 41: 655-666 http://www.intjdevbiol.com/web/paper/9415485

N-cadherin protein distribution in normal embryos and in embryos carrying mutations in the homeobox gene Hoxa-4 A I Packer, V A Elwell, J D Parnass, K A Knudsen and D J Wolgemuth Int. J. Dev. Biol. (1997) 41: 459-468 http://www.intjdevbiol.com/web/paper/9240562

HOX genes and cell adhesion molecules in cancers A Alonso Varona, B Castro, Y Calle, T Palomares, P Bilbao, D Flagiello, M F Poupon and B Malfoy Int. J. Dev. Biol. (1996) 40: S309-S310 http://www.intjdevbiol.com/web/paper/9087811

5 yr ISI Impact Factor (2016) = 2.421